Merge pull request #63 from andrewjpage/master

sliding window at half minimum window size
nickjcroucher · Feb 28, 2013 · bb287fe · bb287fe
2 parents 0e561fd + f97c625
commit bb287fe
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 6 deletions.
diff --git a/src/branch_sequences.c b/src/branch_sequences.c
@@ -356,7 +356,7 @@ void get_likelihood_for_windows(char * child_sequence, int length_of_sequence, i
 		// start at the coordinate of the first snp
 		window_start_coordinate = snp_site_coords[0];
 
-		int number_of_windows = (int) ceil(branch_genome_size/window_size);
+		int number_of_windows = branch_genome_size;
 		// start coordinate, end coordinate, likelihood
 
 		int * block_coordinates[2];
@@ -366,7 +366,7 @@ void get_likelihood_for_windows(char * child_sequence, int length_of_sequence, i
 		block_coordinates[1] = (int *) malloc((number_of_windows+1)*sizeof(int));
 		number_of_blocks = 0;
 
-		for(i = 0; i < ceil(branch_genome_size*1.0/window_size); i++)
+		for(i = 0; i < branch_genome_size; i++)
 		{
 			window_end_coordinate = get_window_end_coordinates_excluding_gaps(window_start_coordinate, window_size, snp_locations, child_sequence,length_of_sequence);	
 
@@ -387,7 +387,8 @@ void get_likelihood_for_windows(char * child_sequence, int length_of_sequence, i
 			}
 
 			int current_window_start_coordinate = window_start_coordinate;
-			window_start_coordinate = window_end_coordinate;
+			// Minimum statistically significant window size divided by 2 (Nyquist).
+			window_start_coordinate += (int) MIN_WINDOW_SIZE/2;
 			// Move to next snp, more efficient but then the adjacent block check doesnt work.
 			//window_start_coordinate = advance_window_start_to_next_snp(window_start_coordinate, snp_site_coords, child_sequence, number_of_branch_snps);s
 

diff --git a/src/gff_file.c b/src/gff_file.c
@@ -29,14 +29,14 @@ void print_gff_header(FILE * gff_file_pointer, int genome_length)
 	fflush(gff_file_pointer);
 }
 
-void print_gff_line(FILE * gff_file_pointer, int start_coordinate, int end_coordinate, int number_of_snps, int current_node_id, int parent_node_id, char * taxon_names)
+void print_gff_line(FILE * gff_file_pointer, int start_coordinate, int end_coordinate, int number_of_snps, char * current_node_id, char * parent_node_id, char * taxon_names)
 {
 	fprintf(gff_file_pointer, "SEQUENCE\tGUBBINS\tCDS\t");
   fprintf(gff_file_pointer, "%d\t",start_coordinate);
   fprintf(gff_file_pointer, "%d\t",end_coordinate);
 	fprintf(gff_file_pointer, "0.000\t.\t0\t");
 
-	fprintf(gff_file_pointer, "node=\"%d->%d\";", parent_node_id, current_node_id );
+	fprintf(gff_file_pointer, "node=\"%s->%s\";", parent_node_id, current_node_id );
 	fprintf(gff_file_pointer, "taxa=\"%s\";", taxon_names);
 	fprintf(gff_file_pointer, "snp_count=\"%d\"", number_of_snps);
 	fprintf(gff_file_pointer, "\n");

diff --git a/src/gff_file.h b/src/gff_file.h
@@ -20,6 +20,6 @@
 #ifndef _GFF_FILE_H_
 #define _GFF_FILE_H_
  void print_gff_header(FILE * gff_file_pointer, int genome_length);
- void print_gff_line(FILE * gff_file_pointer, int start_coordinate, int end_coordinate, int number_of_snps, int current_node_id, int parent_node_id, char * taxon_names);
+ void print_gff_line(FILE * gff_file_pointer, int start_coordinate, int end_coordinate, int number_of_snps, char * current_node_id, char * parent_node_id, char * taxon_names);
 #endif